High dropout rates – Technology to the rescue?

With a high proportion of university students in South Africa dropping out before graduation, many in their first year of study, higher education institutions are turning to technology in an attempt to arrest the declining pass rate. But the complex problem may need a more radical approach.

National graduation statistics point to a university system that is largely failing South African students and, in the long run, society as a whole.

According to the Department of Higher Education and Training or DHET, 24% of all students at universities in South Africa dropped out in their first year of study in 2015; and 14% dropped out in their third year of study. Fifty-two percent of students took seven years to attain an undergraduate qualification, and 48% never graduated.

Struggling to cope, universities are experimenting with high-tech predictive computer programmes to help them identify at-risk students early on.

For several years, the University of KwaZulu-Natal or UKZN has had a 'robot system' in place which, based on students’ academic performance, has identified and classified students at risk of failing. The system classifies students according to the colours of a traffic light, with red indicating a student in danger.

Artificial intelligence

UKZN is now taking the system several notches further with a sophisticated artificial intelligence system which not only analyses students’ performance but offers practical advice and alternatives for them to achieve the credits needed to pass a degree.

Randhir Rawatlal, associate professor and head of research engineering at UKZN, has pioneered the Auto Scholar, or AS, programme as a more systematic and sustainable approach to support struggling students.

Using algorithms, the AS system is able to predict the courses best suited for success, how well a student is likely to perform, and also identify courses which do not serve any foundational role and should therefore not be a prerequisite. An engineering student, for example, may be advised to drop a course in applied linguistics in favour of another more attainable course to achieve his or her credits.

The early indications, based on its roll-out at UKZN’s chemical engineering department, are that the system is effective in at least getting students to stop and consider the advice given. A student who is a recipient of a government grant, for example, and not performing may be reminded by the AS system of his or her obligation.

In extreme cases, the advice may be to drop out of university entirely, but more often the advice centres on alternative courses to achieve credits specific to a student's circumstances and which are needed to attain a degree.

Students often appeared more willing to take advice generated from a computer programme, which they could access on their mobile phones, than from teaching staff, said Rawatlal.

“We find that students are taking the AS advice more seriously and are acting on it,” he said.

Through the cross-correlation of results, the AS system was also able to reveal trends and anomalies.

“If an entire class was performing badly, there should be alarm bells about the lecturer, especially if the students’ marks did not correlate with their high school performance,” said Rawatlal.

‘Overrating’ of maths

An interesting trend that the AS system has picked up is what Rubby Dhunpath, who is a member of the Council on Higher Education’s advisory board and is director of teaching and learning at UKZN, refers to as the “overrating” of maths.

While stressing that further analysis was required before any conclusions could be drawn, Dhunpath said: “The importance of mathematics, which is privileged for entry in just about every university course, is overstated as the AS system has been able to direct students to alternative subjects in order to achieve their credits.”

A similar system to Auto Scholar is used in universities in the United States and has been adopted by the University of Johannesburg, or UJ, for implementation in the second semester of 2017. Blackboard Predict, a product of the Eiffel Corp group which offers technological assistance to education institutions, will be customised to a South African context at UJ.

“Our academic structure is very different to the US, and Blackboard Predict’s fully customisable model perfectly adapts to the SA environment and our specific needs,” said Professor Thea de Wet, director of the Centre for Academic Technologies at UJ, in a statement.

“The roll out of Blackboard Predict will be supported by a tailor-made intervention programme at UJ. The institution operates on a best practice and best evidence model so we will be looking at institutions around the world to gain insight about adding this predictive solution to our already extensive tutoring system and health and psychosocial services."

Socio-economic burden

“If students drop out we not only lose revenue from fees but also from government subsidy; input subsidy received when students start university and output subsidy upon graduation. In addition, there is a heavy socio-economic burden on individuals, families and communities when students do not complete their studies,” said De Wet.

Early detection and interventions are no doubt urgently needed at South African universities but are they mere stop-gap measures to more complex and bigger problems facing universities in multi-faceted and developing countries like South Africa?

Dhunpath said while the systems are useful to identify struggling students and to facilitate support where needed, they were not likely to be effective if the support was not part of a sustained, holistic approach.

Many higher education institutions have for a number of years offered additional writing support, counselling services and other forms of assistance but these reactive measures have not been sufficient to address students’ needs, especially those arising from the gap between high school and university education.

“We assume, for example, that because someone has passed matric they will know how to write an academic article, or that someone who has passed matric-level maths literacy can make the jump into pure mathematics at university,” said Dhunpath.

Support structures

In South Africa, a vast number of students are also first-generation university students, which means they will not necessarily have a support structure outside of the university to help them cope and make decisions. Wider access to government university funding has also seen an increasing number of poor and rural students entering university, for whom university life represents a significant social and cultural shift.

Dhunpath suggested that universities dedicate an entire semester to generic courses for all first-year students in order to acclimatise them to university life. This should be followed by sustained support throughout a student’s stay at university, he added.

But to do so, the “chasm” between academic and support services at universities had to be narrowed to allow for a seamless system providing both academic and other services for students.

“An English lecturer will refer a student struggling with writing to the institution’s writing centre. But does it not make more sense for the lecturer herself to help the student?” asked Dhunpath.

It is an issue the Council on Higher Education or CHE is also attempting to address by the development of an accreditation system for teaching staff at universities. Such a system recognises that there is a need to strengthen pedagogic expertise in higher education.

“We can argue about where the blame lies or we can shift the question to whether we have under-prepared students or under-prepared universities. Universities need to take a critical look at themselves in terms of structure, the systemic nature of higher education curricula, etc, because they have a moral obligation to create the best possible environment to allow students to succeed.”

The right model for students?

At heart is a question about whether the current higher education structure, historically and socially rooted as it is in a centuries-old Eurocentric system, is the right model for students from developing countries like South Africa, said Dhunpath.

“The assumption is that the current model of higher education is the only structure and that we must dismiss any attempt to change it. The third world should deviate from this failed structure and come up with a new structure and curriculum suited to our needs,” he said.

According to Dhunpath, the CHE will focus on curriculum changes in the second phase of its Quality Enhancement Project to interrogate higher education at both institutional level and more broadly.

“We need to interrogate the current curriculum structure which perpetuates age-old practices to determine what exactly is failing our students. For example, in order to pass chemical engineering 1 and 2, you first have to pass maths 1, 2 and 3 but the correlation between the two is not clear,” he said.

Again, these are theories which need to be tested further before any action can be taken. UKZN, however, is in the process of establishing the Institutional Intelligence Research Unit which will, among other functions, assess the large volumes of data emerging from the AS system.

Said Rawatlal: “This system will provide the concrete data needed to determine if there was a need for curriculum and other structural changes.

“There is no simple solution to this very complex problem.”

Note: The article above has been updated with the following correction made: The quote, “The importance of mathematics, which is privileged for entry in just about every university course, is overstated as the AS system has been able to direct students to alternative subjects in order to achieve their credits,” was originally incorrectly attributed to Randhir Rawatlal, associate professor and head of research engineering at UKZN. It should have been attributed to Rubby Dhunpath, director of teaching and learning at UKZN. The error is regretted.